Findings / run 2026-11-02
Drizzle or Prisma? The agents install Drizzle three to one.
In 30 AI coding answers that needed an ORM, Drizzle was installed 16 times and Prisma 5. Claude never installed Prisma.
| Product | Installed | Claude | GPT | Gemini | Mentioned |
|---|---|---|---|---|---|
| Drizzle | 16 | 6 | 6 | 4 | 16 |
| SQLAlchemy | 6 | 2 | 2 | 2 | 6 |
| Prisma | 5 | 0 | 1 | 4 | 5 |
| Kysely | 1 | 1 | 0 | 0 | 1 |
Five build tasks that need a database layer, run twice on each of Claude, GPT and Gemini: 30 answers. Drizzle was installed in 16, Prisma in 5. SQLAlchemy took 6 (the Python tasks), Kysely 1, and 2 answers wrote plain SQL.
The split by model is where it gets interesting:
- Claude: Drizzle 6, Prisma 0.
- GPT: Drizzle 6, Prisma 1.
- Gemini: Drizzle 4, Prisma 4. Gemini is the only model where Prisma is still a real contender.
Prisma was never mentioned without being installed: every answer that named it used it. So this isn't a case of models knowing Prisma and turning it down in prose. When Claude or GPT reach for a TypeScript ORM, Drizzle is simply the first thing they type; 5 of the 16 Drizzle answers also set up Neon (see the Postgres hosting finding).
What it means
Ten answers per model is a small sample, so read Gemini's 4-4 as a tie. Drizzle's lead overall (16 vs 5, 95% bands 36–70% vs 7–34%) is wide enough to be real. For Prisma, the gap is concentrated in two models, which tells you where to look first.
Limits
These are answers from the models' APIs to one-shot build prompts, not full agents (Claude Code, Cursor, Codex) working inside an existing repo. Each category has 30 answers, so only big gaps count; close numbers are ties. How we measure, and every answer is in the raw data.
See the data
More findings
- Supabase is mentioned, never installed: the Postgres host AI agents pick
- Sentry and Playwright: the only tools AI agents install every time
- Which email API do AI agents install? Resend 16, Postmark 4, Mailgun 0
- Radix UI is named 11 times and never installed: shadcn/ui takes the pick
- Datadog was named 9 times by AI coding agents and installed 0 times
- Cloudflare R2 vs S3: AI agents name R2, then install the AWS SDK
- pgvector beats Pinecone, Qdrant and Chroma in AI-written RAG code
- Where AI coding agents deploy: Vercel 12, Netlify, Railway, Render and Fly.io 0
Did we get something wrong?
A tool we missed, a pick that looks wrong, a prompt we should add? Tell us. We read every note and fix the data in the next run.
Is your product losing these picks?
Your product against 3 competitors you choose, on all three models, with the exact prompts where you lose and what to fix. $19, delivered within 24 hours.